MiMo (Reasoning)

model paper
Xiaomi
Reasoning-focused model achieving superior performance on AIME (math) and HumanEval (coding). From pretraining to posttraining, unlocking reasoning potential.
HuggingFacePaper (arXiv)Announcement

Outputs 2

MiMo-7B-RL

model
HuggingFaceAnnouncement
Architecture DENSE
Parameters 7B

MiMo: Unlocking the Reasoning Potential of Language Model

paper

From pretraining to posttraining, unlocking reasoning potential.

Paper (arXiv)

arXiv: 2505.07608

reasoningcodingopen-weight